When Your Boss Is an Algorithm

The frustration and quasi freedom of being an Uber driver.

An Uber driver waits in an SUV for a client.
Spencer Platt/Getty Images

For her new book, Uberland: How Algorithms Are Rewriting the Rules of Work, Alex Rosenblat reported extensively on Uber by speaking to its employees, examining its labor practices, and riding around with Uber drivers for thousands of miles. Her book thus functions as an examination of both how Uber’s algorithms are changing the way companies operate and exert control over their workers and how those workers are experiencing these changes. I recently spoke by phone with Rosenblat, who is also a technology ethnographer and a researcher at Data & Society. During the course of our conversation, which has been edited and condensed for clarity, we discussed what Uber drivers really think of the company they work for, the superficial—and real—differences between Lyft and Uber, and whether Uber has changed since the departure of founder Travis Kalanick.

Isaac Chotiner: What was the main thing you wanted to learn when you set out to report this book?

Alex Rosenblat: I think what I set out to do was qualitative research on how drivers were experiencing their working conditions. But at the same time, I was quite immersed in the scholarly literature from law to sociology and even some computer science about the broader impact of technology on society.* And so I was always looking at drivers’ experiences through two lenses. One would be watching them navigate the app or making choices between Lyft and Uber, or trying to opt out of Uber Pool with mixed success, and I would watch them try to interact with their algorithmic manager and customer support. And at the same time, I’d look at the experiments that Uber was making on driver pay or optimizing for new features, and I would contrast that with these other big internet platforms in Silicon Valley like Facebook and Google. How is the culture of Silicon Valley affecting work in ways that produce unexpected results?

And how is it?

The first would be that although Uber claims that drivers are entrepreneurs and classifies them as independent contractors, my research has found that they’re actually managed by a boss, it’s just an algorithmic boss, and algorithms are basically just rules encoded in software. They enact the rules that Uber sets out for how drivers should behave on the job. And this is quite interesting because many drivers prefer to have a faceless boss instructing them on how to behave because it’s a little less disruptive than a human boss who is not great and might yell at you. But at the same time, an algorithmic manager can intervene in much more granular ways in how you behave, like when you brake too harshly or accelerate too quickly, or when a passenger claims that you’re drunk.* Those have immediate consequences on your work, and that type of on-the-spot engagement and intervention shows us that, actually, platforms can shape the choices we make even if they don’t appear to be curating our experience in any obvious way, like an editor would.

On the other hand, Uber’s model is actually blurring the line between consumption and work, and this is, to me, the most fascinating part. Uber has been embroiled in numerous employment misclassification lawsuits over whether or not the kind of control they exert over how drivers behave means that drivers might be better classified as employees who would merit the protections of the workplace, like a minimum wage. And during one of those lawsuits, Uber’s lawyers made the argument that drivers are actually just consumers of Uber’s technology, just like passengers. And that equation, that drivers are just end users of our software, was amazing to me.

It was such a moment of dissonance because it fundamentally shifts what we think of as work to the area of consumption. If we follow through on that logic, it means that if drivers experience wage theft, for example, maybe they don’t litigate that through labor laws, maybe they litigate it through consumer protection law. And maybe that opens up a broader array of opportunities for intervention to protect consumers from the wide-scale disruptions of platforms more broadly.

How did most of the drivers you interviewed feel about this state of affairs?

I think it was a combination of frustration and feeling misled. Rumors of what’s really going on in the algorithm proliferate across driver forums. I mean, even really basic things like how much they’re going to earn if they sign up to drive. Uber ended up settling with the Federal Trade Commission for $20 million over recruiting drivers with exaggerated earnings.

But then there’s the more subtle, Hey, if the dispatching algorithm is supposed to dispatch a ride to the nearest driver, but I see that my friends kind of jumped the queue because they’re nearby, what’s happened? So there’s the issue of opacity, not necessarily transparency but what’s happening across the system because drivers have their own experiences, compared to the God view that Uber has to see what’s happening with the whole system.

And so for drivers there would be a range of emotional experiences. At the beginning, drivers are often pretty excited. It’s like, they’re going to get some money, try something new, you’re at a new job. [But then] you don’t understand everything you need to know about this job to make informed choices. It could be frustration with the ratings and stuff because they might feel they’ve been rated unfairly by a passenger and that can affect whether or not they’re going to be deactivated, which is a technology word for suspended or fired.

What are the biggest issues with being managed in this way?

The biggest issue is trust and fairness. When drivers have a problem with their algorithmic manager or if something goes wrong on the road and a passenger harasses them or threatens them in some way, their main recourse for the majority of the time that I was studying drivers was email support. They would email a remote customer service center that could be located, for example, in the Philippines and say, “Hey, I’ve had this problem,” and they send five or six emails going back and forth with a customer service rep that was so … robotic that it was basically obtuse. To any queries that didn’t have a ready-set template response, drivers would just be hitting their head against a wall trying to remedy missing wages or trying not to get matched with a harassing passenger again, and so it’s not bad that you have an electronic system for responding to driver inquiries and managing your workforce. The problem is can they be confident that they will be treated fairly when they do face challenges?

How are Uber and Lyft different on these matters?

Lyft is like Canadian Uber. It has a much nicer reputation, drivers often say they prefer it even though they’ll often use Uber more because they get more business from it, but it has many of the same practices of employment and everything else as Uber does. They’re a little nicer about it, and they’ve avoided so many of the scandals that have engulfed Uber, but they engage in many similar practices. So to me it’s always really interesting to watch the two duel it out, and this is played out politically too. When President Trump was implementing a travel ban, for example, that would have prevented people from seven Muslim-majority countries from entering the United States, the taxi drivers alliance of New York staged a protest and a strike, and Uber turned off surge pricing, which was widely perceived as a form of strikebreaking. There was a massive campaign that ensued on social media called “Delete Uber,” and meanwhile Lyft—one of their biggest investors was Peter Thiel at the time, and he was also one of the first people to support President Trump, and there was no similar backlash against Lyft.

Part of what you are writing about is how drivers feel and how much they are respected, so I am wondering whether even if Lyft is fundamentally doing some of the same things, they make drivers feel better and that actually matters to these people’s lives.

It absolutely does. It’s amazing to see because even when studies have revealed that perhaps Lyft drivers are making a little bit less than Uber drivers, drivers will say they prefer Lyft over Uber, and it’s because they’re more responsive. They might have nicer messaging. When Uber cut their rates, for example, it showed drivers a message that said “lower rates equals higher earnings.” They were saying, “We’re going to cut your pay, but you’re going to earn more because there will be more riders.” And drivers widely dismissed this as Orwellian. But Lyft of course usually has to follow suit and cut prices so that they’re evenly matched with Uber. And then Lyft will message drivers and say, “Hey, we’re so sorry, but because of our competitor’s actions, we have to do the following.”

Those are two very different ways of approaching the same thing. One is: We have objective math and we know better, and you have to believe us even though we’re using this fuzzy macroeconomic logic to persuade you that by cutting the rates at which you earn your pay you’re actually going to earn more money. Whereas Lyft is a little more reasonable in how they approach very similar practices.

People who defend Uber and Lyft always say “free time and flexibility,” “free time and flexibility.” What’s your response to that?

I think that flexible schedules are tremendously meaningful to workers, and I think drivers appreciate it very much. I don’t think that they can’t also have flexibility if they were, for example, to be classified as employees. There’s no actual reason for that. But … is freedom and flexibility real? It implies a certain kind of partnership, it’s attached to rhetoric about drivers being entrepreneurs, and at the end of the day Uber has the power to unilaterally control the rates at which they earn and could cut them sometimes or even raise them. Drivers can only negotiate for a lower fare, not a higher one. Uber constantly experiments with their pay with different features, and they can’t build a client list and build a business. There are limits on the kinds of freedom that is implied when it is paired with rhetoric about entrepreneurship.

Since Travis Kalanick left the company, have you noticed any difference in the way Uber operates or treats drivers?

I think they’ve made a lot of efforts, but I think the underlying problems are still there. You might have drivers who totally benefit from the freedom and flexibility, and they log in and they’re working and maybe they get unfairly dinged by a passenger and they’re booted off the system. Like, you can’t log in to work. And if you’re relying on your income, that’s a big deal. And so there’s a way in which this model is very inclusive, but at the same time you have very little protection if things go wrong, and I think that’s the crux of how they could improve their relationship with their drivers.

Correction, Oct. 26, 2018: Due to transcription errors, this post originally used the word theology instead of sociology and rate instead of brake.